New-less JavaScript (2013) | Hacker News Reader

New-less JavaScript (2013)

jfire.io

37 pointsvoltagex_43 comments

I use `delete object[key]` all the time to unset a property. I do think in the JS world the use of `new` is inconsistent, but avoiding the keyword does not rid of that fact. Your new-less libraries may be consistent with each other, but not the rest of the JS world that do it their own way. (http://xkcd.com/927/)

This seems like a problem the language should solve incrementally. Especially with ES6 classes and modules!

`delete` is harmful, consider setting property to `undefined` or `null` instead.

Why is it harmful to delete a property that disappears forever?

It is going to depend partly on the internals of your JIT. Since JS objects are so mutable it's common to use their shape as the basic check for fast method dispatch. So if you take a Foo and add a property x then it will become a Foo(x). If you make lots of calls on objects of this shape then the method lookups will make it into inline caches and be very quick, calls against an object of shape Foo() or Foo(y) or Foo(x,y) will be treated separately. Inline caches for method dispatch are generally very small, so you'll omly get really fast dispatch for a few different shapes at any particular call site so the more consistent the shapes the more likely you are to keep the JIT happy.

So your argument is that we should currently and in the future constraint our usage of language-features based on the current state of language-runtimes?

How are we going to keep pushing the boundaries then? Had JS started with that sentiment, we would have a V8 JS runtime running about 100th the speed it does now for general purpose code, written with the aim of being as clear and reusable as possible.

Personally I prefer good, clear and concise code, over code written trying to appease some secondary affects on the inside of a black box which is not guaranteed to stay the same.

This is clearly an optimization, and I'm not saying all optimization is bad per se, but all optimization should be justified if it affects code-clarity. Are you sure these optimizations are warranted? Have you profiled your code and identified that this is a bottleneck for performance-critical sections of your code?

A former Nvidia-engineer had a really good rant about what sort of things your thinking leads to:

http://www.gamedev.net/topic/666419-what-are-your-opinions-o...

Key quote: "So the game is guessing what the driver is doing, the driver is guessing what the game is doing, and the whole mess could be avoided if the drivers just wouldn't work so hard trying to protect us."

I agree it's not a direct analogue, but the same line of thinking still applies.

My argument is that you should be aware of what is required to implement a language (at least to some extent), and what may go against implementation assumptions. These things are always a tradeoff, and if you mutate objects enough through their lifespan then maybe implementations will move more towards looking at the methods attached to an object, and ignore the shape of the data except when accessing a property (the methods are already normally stored off to one side on the assumption that you won't change them), but that will either make objects larger (need to store a ref to the shape and to the methods per object) or require an extra indirection (need to follow from the shape to the methods) so may incur a slight overall performance penalty.

Yes, we are all in the same game as GL driver developers and game devs, and that goes for all language implementations. The fine details of what gets optimised best may change from VM to VM, and from release to release, but it's important to understand the general assumptions that these systems are built around, because they don't change quickly.

In general dynamic languages that allow properties to be created and deleted on individual objects are a complete PITA to optimise, and more generally mutability of classes makes stuff harder, so in general even though you have these abilities in JS and other languages your code will often perform better if you limit your use of them, especially in areas where performance is critical. This doesn't mean you shouldn't use those mutability features, but you should have some notion of the cost you might be incurring.

Thanks. That explained it really well. Semantically it makes sense to me, but I didn't take the interpreter into account.

Because garbage collectors doesn't care if it is simply deleted. Setting it to undefined or null is better for speed

http://bertanguven.com/preventing-memory-leaks-in-javascript...

This article is just badly written. It doesn't provide or make reference to a precise documentation of the issue. It's not clear whenever it applies to `delete x` or also `delete obj[x]`. It doesn't provide a reproductible use case or list of browsers that are affected. It mentions issues with scope but doesn't go in detail of which ones are triggering the issue. For what it's worth it might apply only to IE6.

`delete obj[x]` removes not only the reference to the value but also the key which prevents other forms of memory leak.

There is another one from "Smashing Magazine" [1], I think it's written better :

> In quite a few discussions online about reclaiming memory in JavaScript, the delete keyword is brought up, as although it was supposed to be used for just removing keys from a map, some developers think you can force de-referencing using it. Avoid using delete if you can. In the below example, delete o.x does a lot more harm than good behind the scenes, as it changes o‘s hidden class and makes it a generic slow object.

Although it's from 3 years ago it looks decent. The comments below are even more interesting than the article.

[1] : http://www.smashingmagazine.com/2012/11/05/writing-fast-memo...

Thanks. Went for `undefined` because it's for query string building, where `null` still is a valid key-value combination.

One of the greatest things about preprocessing languages is that they usually use optimized and hard-tested features of the specific language.

Let's examine CoffeeScript ( which is one of the most popular ) :

    class Foo
      constructor: -> return 'bar';

    foo = new Foo();

will compile to :

    var Foo, foo;

    Foo = (function() {
      function Foo() {
        return 'bar';
      }

      return Foo;

    })();

    foo = new Foo();

Which IMHO, follows the very best practices of javascript. Since JS is such a flexible language if you don't have any coding standards, your code could become hell and unmaintainable very quickly.

I would advise not to use the proposed "New-less JS", because it breaks the general coding style guides that a lot of libraries use and yes it would be possible to write it that way, but keep in mind that your code would be one step less readable from random developers.

Is that really optimized/optimal? I'd be curious to see jsperf benchmarks of this method of constructor vs the straighforward approach.

The JS VMs tend to kind of finnicky and optimized towards certain types of idiomatic Javascript and it's my guess that the further off the common path you stray, the more likely you are to hit performance gotchas.

Your best bet is to transpile code that looks much like the benchmark suites they run. :)

I also agree with what you are saying. JS in terms of usage is so big that I guess developers behind JS compilers will have to optimize for "the most popular way of solving that problem".

In any case, why I mentioned preprocessors is that there is a lot of history behind every feature. As of this there are a couple of github issues [1] that explain why this approach has been used.

For the performance. Seems that "new" is way faster. [2]

[1] : https://github.com/jashkenas/coffeescript/issues/912 [2] : https://jsperf.com/obj-create-vs-new/4

If you want your code to be easily readable by as many developers as possible you should probably stay away from CoffeScript.

That's true, but for many reasons it helps, when you are in doubt for a way to do something. I usually use a lot of CoffeeScript's ideas. Namely class inheritance.

Of course when I work in a team or on a big project I don't even consider it, because of future developers and the Bus factor.

"JavaScript borrowed the new and delete keywords from its less-dynamic predecessor languages. They feel a bit out of place in a garbage collected language and are a source of confusion for newbies "

They are? Delete maybe, due to language symantics - but not new. I think it's unfair to bundle them together in terms of newbie confusion.

Also every time I see one of these type of blog post, discussions in a book the work arounds are even worse.

JS is easy to understand, by the time you need 'new' it is not that difficult an extension of your current knowledge. As far as 'delete' even an advanced JS developer could go a year without calling it

'new' is a source of confusion for newbies and experienced devs alike, sufficiently enough for them to ask questions on SO: http://stackoverflow.com/questions/9468055/what-does-new-in-...

Even I have to refresh my memory on these ambiguous parts of the language from time to time.

Fully agree.

The author probably never used Smalltalk, just to cite one of the oldest dynamic languages.

    btn := Button new.

The nice thing about JS is that it has so many "religions". But some people think there should only be ONE religion. While One religion would be easier for newbies, destroying all other religions would kill the language.

So can we just agree that many religions are nice. It's OK to try and convert people, but it's Not OK to force someone by changing the language itself.

With that said, I'm a bit worried about more and more religions entering the JS language. Like classes and block scope (ES6). And how some, both good and bad new features are forever changing the language.

I think a good complement tothis concept is that of "this"-less Javascript, relegating it only to things like mixins.

So many headaches are to be had if you're not careful and using some decent functional things.

(author here)

Absolutely. My preference for new modules is to use both newless and "this"-less Javascript. Newless via the technique outlined in this post (or encapsulated in a module: https://github.com/Mr0grog/newless), and this-less via the module pattern (see link in post).

This is the pattern followed by D3. It's easy to follow internally, gives you true private data and methods, and (IMHO) is a nicer interface for API consumers -- no `new`, `delete`, or `bind`'s necessary.

With ES6 classes everywhere soon , I doubt you'll see less "new". On the contrary, you'll see more java-ish code and less Doug Crockford like code. By the way, he aint in TC39 anymore. So expect even more java-ish features in ES.

I don't think there's anything inherently evil, bad, or even particularly confusing about this. As long as you understand how it's being attached to a running function there's very little confusion to be had. The problem is that most of the resources for learning completely gloss over this incredibly important aspect of the language.

...and too few of the posts complaining about it mention strict mode or ES6 fat arrow syntax.

The problem with module pattern which enables you to use "this"-less Javascript is increased memory usage. Of course you can use other techniques for elements which you instaniate in thousands, but then you are mixing one pattern with another and it will begin hard to follow which should be which. I prefer consistency here.

In theory, an optimizing VM/JIT should be able to handle this for you. I'm not sure if existing implementations currently do any optimizations though.

Agreed, `new` and `this` are pretty much useless boilerplate noise. Gimme a closure and watch me wipe out a quarter of the text in your source and never even have to think about using a`bind`.

Yes, absolutely. In fact, this post is really just providing a trick for not having to always write `new`. Using closures for object construction is the real 'new-less' JavaScript. As a nice parallel, the trick explained in the post lets you construct objects using `new` without typing it, whereas object factories that use closures allow you to use `new` without requiring it.

The library StampIt uses factory methods instead of new to allow some extra features. Works a lot like .extend() in other libraries but with some extra options.

https://github.com/ericelliott/stampit

It's a movement in wrong direction. It's artifact that we can call class name without "new" and method name, because before ES6 there wasn't such thing as classes. Now we should just deprecate this artifact.

The best reason to have new-less constructors is because it's now possible to use them as higher-order functions.

    [1,2,3].map(User)

As long as the constructor doesn't take multiple arguments [1], otherwise:

    [1, 2, 3].map(id => User(id))

[1] http://speakingjs.com/es5/ch15.html#_pitfall_unexpected_opti...

I find it useful to have `unary` and `apply` around:

  [1,2,3].map(unary(User))
  [[1,2],[3,4],[5,6]].map(apply(User))

Why JavaScript, why !

Or the alternative :

    [1,2,3].map( function(i) { return new User(i); } );

might seem more reasonable in the JS world.

It's a valuable but unnecessary intermediate step if you go into functional idioms (map, fold, closures) fully.

raganwald books cover this way of using javascript https://leanpub.com/u/raganwald

How is this news? It says 2013 in the title ...

I feel like being bullied for having to learn Javascript to do web programming. I wish there was a push towards supporting arbitrary languages for doing web development.

Well, that's possible with compile-to-JavaScript languages. Also you have the option of plugins like Unity, if you're going for something graphically adventurous.

That being said, JavaScript is a powerful tool, if wielded correctly. It's so flexible. In a lot of ways, it's like a pizza. One man orders a pizza with anchovies, another with green peppers. And on and on. Eventually it's like they've got two completely different dinners.

> Well, that's possible with compile-to-JavaScript languages.

I have tried to read code that was translated to Javascript using some source-to-source compilers and they are very hard to understand. On the other hand for a language like C, it is very easy to understand the output ( assembly ). Don't get me wrong, understanding the intricacies of assembly for each processor architecture is a difficult process, but the rules that define assembly language itself are not so, there is an opcode and some operands and each instruction performs an action on the operands or stored state in various registers. Javascript ( at least for me ) is too high-level to attempt to read a machine translated output. Programmers need to know at least one level below the abstraction to which they are programming to.

I would say the output of most of the compile-to-js languages are easy or at least non-hard to read. It is usually the minification that causes problems and you don't have to do that in a development environment where you usually attach your debugger. It's usually also hard to debug a running application written in a static and compiled language (unless you have symbol tables and similar).

It's not like pizza, it's like dirt. You can shape it many ways but it tends to fall apart.

Reply on news.ycombinator.com